Exploring the Reliability Features of Linkerd

In this chapter, we will go through Linkerd's reliability features, such as automatic load balancing through dynamic requests routing, service profiles, retries, timeouts, and proactive error code hunting.

This chapter will explain the traffic management capabilities of a service mesh. Understanding this framework is necessary if you want to use networking resources that coordinate, restrict, and scale traffic throughout microservices.

We will be covering the following topics in this chapter:

• Load balancing traffic between services in the booksapp application
• Understanding and creating service profiles
• Routing traffic between the booksapp application for a newly created service profile
• Deploying service profiles for the booksapp application using Swagger specs
• Understanding aggregated routing for the booksapp application
• Enabling...

Technical requirements

This chapter is dependent on Chapter 15, Installing Linkerd. You will need to complete the hands-on exercises of Chapter 15, Installing Linkerd, that deal with the following, in order to work through this chapter:

• Setting up Linkerd
• Installing a control plane
• Deploying the booksapp and emojivoto applications

Make sure that you are in the ~/linkerd/scripts directory for the exercises in this chapter:

$ cd ~/linkerd/scripts

Now, we're ready to begin!

Overview of the reliability of Linkerd

Reliability is a critical quality of any enterprise application. It is even more significant if the application is deployed in an ever-changing environment that meets business requirements through software innovation. Ibryam defines this concept as follows:

• Fragile: The system is unable to survive under stress.
• Robust: The system withstands stress to an extent, and then it breaks.
• Resilient: The system adapts to stress and failures before it reaches the breaking point.
• Anti-fragile: It feeds on stress and change and so it's much harder to create.

This is better understood with the help of the following diagram:

In parallel to reliability, a service mesh provides a resilient system through continuous improvement from sidecar proxies.

The use of specialized libraries such as Twitter's Finagle and Netflix's Hystrix help...

Summary

Applications can switch from being robust to resilient through the use of Linkerd's sidecar proxy, which provides adaptive load balancing, easy to understand debugging capabilities, timeouts, and retries. We explored each of these capabilities in this chapter.

With the help of a service profile through Kubernetes' custom resource definition, you can define routes to report aggregated metrics on unique requests based on defined patterns. The service profile name is a fully qualified name that can match with HTTP/2 :authority or HTTP1.X hosts. Linkerd's load balancing implementation is at the L7 (application streams) level instead of the default Kubernetes L4 (TCP connection) level. You can implement retry budgets to prevent retry storms from overwhelming backends.

The Linkerd dashboard or the Linkerd CLI can be used to observe the live traffic arriving in...

Questions

1. Kubernetes does load balancing at the connection level (L4).

A) True
B)False

2. Linkerd does load balancing at the application level (L7).

A) True
B)False

3. Linkerd load balancing is out of the box, and it requires no special configuration.

A) True
B)False

4. A Linkerd retry requires configuration.

A) True
B)False

5. Linkerd's service profile can be generated automatically, even if the Swagger API is not available for the service

A) True
B)False

6. The retry budget is about done for adaptive retries instead of a fixed number of retries.

A) True
B)False

7. The service profile is needed to provide aggregate route metrics, especially if a parameter is part of the route, which makes it difficult to group a route path.

A) True
B)False

Further reading

• From Fragile To Antifragile Software – Red Hat Developer Blog, Ibryam, Bilgin, Red Hat Developer Blog, 2019: https://developers.redhat.com/blog/2016/07/20/from-fragile-to-antifragile-software/
• Linkerd v2: How Lessons from Production Adoption Resulted in a Rewrite of the Service Mesh, Morgan, W. (2019), InfoQ, available at https://www.infoq.com/articles/linkerd-v2-production-adoption, accessed 5 May 2019
• gRPC Load Balancing on Kubernetes without Tears, Morgan, W. (2018), Linkerd.io, available at https://linkerd.io/2018/11/14/grpc-load-balancing-on-kubernetes-without-tears/, accessed 5 May 2019
• What is L4-L7 Network Services? Definition and Related FAQs | Avi Networks, Avi Networks, (2019), available at https://avinetworks.com/glossary/l4-l7-network-services/, accessed 6 May 2019

• Linkerd 2.x With Network Policy, Sim, I. (2019), available at https://medium...